Epidemiologic studies that seek to evaluate the effects of environmental exposures on human reproduction require new statistical approaches. Urgent tasks are (1) to provide methods of deriving statistical power in studies using multiway contingency table analysis and (2) to consider whether and how studies on reproductive outcomes may utilize repeated pregnancies occurring to the same couple, as opposed to utilizing only a single index pregnancy for each couple. Earlier work based on a study of spontaneous abortion provides the data on which solution to these methodological problems will be evaluated. Preliminary statistical approaches are presented, including an elaboration of the mathematical constructs that have so far been developed.